The Hash implementation assumed that there were always empty buckets, but
sometimes there were only active or deleted buckets (no empty buckets).
Therefore, fix it so that this situation does not occur.

### Example

```ruby
# example.rb
class A
  attr_reader :v
  def initialize(v) @v = v end
  def ==(o) @v == o.v end
  def hash; @v end
  def to_s; "#{self.class}[#{@v}]" end
  alias eql? ==
  alias inspect to_s
end

keys = (0..31).map{A.new(_1)}
h = {}
(0..16).each{h[keys[_1]] = _1}
(17..31).each do
  k = keys[_1]
  h[k] = _1
  h.delete(k)
end
p h.keys
```

#### Before this patch:

```console
$ bin/mruby example.rb
[A[0], A[1], A[2], A[3], A[4], A[5], A[6], A[7], A[8], A[9], A[10], A[11], A[12], A[13], A[14], A[15], A[16], A[30], A[31]]
```

#### After this patch:

```console
$ bin/mruby example.rb
[A[0], A[1], A[2], A[3], A[4], A[5], A[6], A[7], A[8], A[9], A[10], A[11], A[12], A[13], A[14], A[15], A[16]]
```